COPI is essential for Golgi cisternal maturation and dynamics

Visualization of cisternal maturation using transmembrane Golgi-resident proteins

We next examined cisternal maturation of the Golgi in detail by using these Golgi-resident markers. We conducted simultaneous dual-color 4D observations at time resolution of 8 frames/s by SCLIM (Kurokawa et al., 2013; Matsuura-Tokita et al., 2006). Similar to the case in the cell expressing mRFP–Sed5 and Sec7–GFP (Fig. 2C and Movie 3), early cisternae labeled with Mnn9–mCherry changed their color to those of medial cisternae labeled with Gnt1–GFP or trans cisternae and TGN labeled with Sys1–GFP, giving evidence for cisternal maturation through recycling of Golgi-resident transmembrane proteins (Fig. 2A,B; Movies 1 and 2). We determined the transition period between the intensity peaks for each pair of fluorescent markers on a single cisterna (peak-to-peak time) (Daboussi et al., 2012). The average peak-to-peak time from Mnn9–mCherry (cis) to Gnt1–GFP (medial) (38.6±11.6 s) was shorter than that from Mnn9–mCherry (cis) to Sys1–GFP (trans) (45.6±19.6 s) (mean±s.d., n>11; Fig. 2A,B). The transition period from mRFP–Sed5 (cis) and Sec7–GFP (trans) (87.8±34.0 s) was longer than that from Mnn9–mCherry (cis) to Sys1–GFP (trans) (Fig. 2C). Even though experimental variability was high, these results were consistent with the spatially ordered distribution of Golgi-resident proteins within the Golgi. High-resolution images during the transition of two fluorescent signals revealed that later markers of the Golgi cisternae, Gnt1–GFP, Sys1–GFP and Sec7–GFP, began to accumulate as small punctate structures on the early cisternae and increased their volume to cover entire regions on the membrane. By contrast, the earlier markers, Mnn9–mCherry and mRFP–Sed5 signals showed a gradual reduction (Fig. 2A–C). In addition, we noticed that earlier and later Golgi markers did not completely overlap but rather showed segregation within the maturing cisternae, corroborating that functional domains exist within a Golgi cisterna (see also Fig. 1C). We could not observe transition from Mnn9–mCherry-labeled cis cisternae to Sec7–GFP-labeled trans cisternae, indicating that Sec7–GFP begins to accumulate later than the disappearance of Mnn9–mCherry from cisternae. This is consistent with the above finding that Mnn9 rarely colocalized with Sec7 (Fig. 1A,B; Fig. S1). The two trans markers Sys1 and Sec7 probably accumulate on cisternae at different maturation phases; Sec7 comes later than Sys1.

Open in a separate window

Fig. 2.

4D observation of cisternal maturation. Wild-type cells expressing (A) Mnn9–mCherry (cis, magenta) and Gnt1–GFP (medial, green), (B) Mnn9–mCherry (cis, magenta) and Sys1–GFP (trans, green) and (C) mRFP–Sed5 (cis, magenta) and Sec7–GFP (trans, green) were grown to a mid-logarithmic phase in synthetic medium at 30°C and observed by SCLIM. Upper left panels show representative 3D images. Dashed lines indicate the edge of the cells. Upper right montages show 3D time-lapse images (4D) of the indicated squares. Lower left panels show the relative fluorescence intensities of green and red channels in the indicated cisterna. Lower right panels show magnified images of cells at selected time points. Scale bars: 1 µm. Representative images from at least 26 independent cisternae are shown.

4D observation of cisternal maturation in temperature-sensitive COPI mutant cells

In our previous report, we showed that cisternal maturation was slowed down but was still observed in α-COP temperature-sensitive mutant ret1-1 cells at the restrictive temperature (Matsuura-Tokita et al., 2006). Because this previous observation was made only in 2D time lapse, we decided to reinvestigate this issue using our much improved 4D imaging system, SCLIM, which has high sensitivity and high resolution.

Both in wild-type and ret1-1 cells at 25°C, transition of the fluorescent signals from mRFP–Sed5 to Sec7–GFP was observed within 120 s, indicating that efficient cisternal maturation occurred in the cells (Fig. 3A,B). By contrast, in ret1-1 cells at the restrictive temperature, cis cisternae labeled with mRFP–Sed5 did not lose red fluorescence but kept it for more than 400 s, and never acquired Sec7–GFP signals (Fig. 3B; Movie 4). The rate of successful Golgi maturation was greatly decreased (11%, 5 out of 55 cisternae) in ret1-1 cells at the restrictive temperature compared to the permissive temperature (62%, 13 out of 21 cisternae). No significant difference was observed for the wild-type upon temperature shift (Fig. 3C). These observations indicate that COPI function is indispensable for cisternal maturation.

Open in a separate window

Fig. 3.

4D observation reveals the defect of cisternal maturation in the α-COP mutant ret1-1 at the restrictive temperature. Wild-type (WT) and ret1-1 cells expressing mRFP–Sed5 (cis, magenta) and Sec7–GFP (trans, green) were grown to a mid-logarithmic phase in synthetic medium at 25°C. Wild-type cells cultured at 25°C and 38°C for 10 min (A) and ret1-1 cells cultured at 25°C (permissive) and 38°C (restrictive) temperature for 10 min (B) were observed by SCLIM. Representative 3D images are shown. Dashed lines indicate the edge of the cells. Right montages show 3D time-lapse (4D) images of the indicated squares. Scale bars: 1 µm. (C) The numbers of cisternae that had matured are shown. At least independent 17 cisternae were counted.

4D observation of cisternal maturation in cells depleted for COPI proteins

To further elucidate the role of COPI proteins in cisternal maturation, we developed cells in which COPI proteins were degraded through an auxin-inducible degron (AID) system. The AID system allows specific protein degradation in the presence of plant hormone auxin in non-plant cells (Nishimura et al., 2009). Arabidopsis thaliana E3 ubiquitin ligase TIR1 was constitutively expressed in yeast cells and the A. thaliana IAA17 sequence was added as a tag at the C-terminus of Ret1 (α-COP) or Sec21 (γ-COP). These cells are designated Ret1-aid and Sec21-aid cells, respectively, or COPI-aid cells, collectively. Expression of IAA17-tagged Ret1 and Sec21 could rescue the growth defects of corresponding temperature-sensitive mutants at the restrictive temperature (Fig. 4A). COPI-aid cells exhibited strong growth defects in the presence of 0.5 mM 1-naphthaleneacetic acid (NAA, a synthetic auxin), whereas they grew normally as wild-type cells in the absence of NAA (Fig. 4B). More than 80% of IAA17-tagged COPI proteins were degraded upon incubation with 1 mM NAA for 2 h (Fig. 4D). Auxin-dependent degradation of COPI proteins was reversible, because the growth rates of COPI-aid cells after NAA washout were comparable to that of wild-type cells (Fig. 4C). ER and Golgi precursor forms of a vacuolar protein, carboxypeptidase Y (CPY), accumulated in COPI-aid cells after NAA treatment for 2 h (Fig. 4D).

Open in a separate window

Fig. 4.

Degradation of the COPI proteins Ret1 and Sec21 by the AID system. (A) Expression of IAA17-tagged Ret1 and Sec21 from low-copy plasmids rescued the temperature-sensitive phenotypes of ret1-1 and sec21-1 or sec21-2 cells, respectively. (B) Growth of wild-type (WT), AtTIR1, Ret1-IAA17, Ret1-aid, Sec21-IAA17 and Sec21-aid cells on a YPD plate with or without 0.5 mM NAA at 30°C. (C) Growth of NAA-treated cells after NAA washout. Wild-type, Ret1-aid and Sec21-aid cells were cultured in YPD medium with or without 1 mM NAA for 2 h at 30°C. After washing out medium with water three times, cells were grown on a YPD plate at 30°C. (D) Ret1-aid and Sec21-aid cells were cultured in YPD medium with or without 1 mM NAA at 30°C. Cells were collected at 0, 1 and 2 h after NAA addition. Ret1-IAA17, Sec21-IAA17 and CPY proteins were detected by immunoblotting. The positions of the ER form (p1), Golgi form (p2) and mature (m) CPY are indicated on the right. The PGK lane shows the loading control. All experiments were performed three times and a representative example shown.

Using COPI-aid cells expressing Mnn9–mCherry and Gnt1–GFP, we next examined the effect of COPI protein removal on cisternal maturation. In COPI-aid cells treated with 1 mM NAA for 2 h, cisternae with the signal of Mnn9-mCherry were sustained and never acquired the Gnt1-GFP signal (Fig. 5A–C; Movies 5 and 6). The rates of cisternal maturation in COPI-aid cells were significantly reduced by NAA treatment (Fig. 5F). These observations again indicate that COPI function is required for the retrograde flux of Golgi-resident proteins and is essential for cisternal maturation of the Golgi. Intriguingly, functional domains labeled with Mnn9–mCherry and Gnt1–GFP were kept segregated within the cisterna in NAA-treated COPI-aid cells (Fig. 5D,E), suggesting that COPI function is also necessary for domain dynamics. We next examined the effect of COPI depletion on the behaviors of the trans-Golgi and TGN marker Sys1–GFP. Transition from Mnn9–mCherry to Sys1–GFP was also disturbed in the presence of NAA in COPI-aid cells (Fig. 6). Cis cisternae labeled with Mnn9–mCherry never lost their color (Fig. 6A,B, boxed area 2; Movies 7 and 8). Cisternae harboring both Mnn9–mCherry and Sys1–GFP sustained both fluorescent signals for a prolonged time (Fig. 6A,B, boxed area 1; Movies 7 and 8). These results indicate that the function of COPI is required over the whole maturation process of cis-to-trans cisternae within the Golgi.

Open in a separate window

Fig. 5.

Depletion of COPI protein inhibits cis to medial cisternal maturation. (A) Wild-type (WT), (B) Ret1-aid and (C) Sec21-aid cells expressing Mnn9–mCherry (cis, magenta) and Gnt1–GFP (medial, green) were grown to a mid-logarithmic phase in synthetic medium with or without 1 mM NAA at 30°C and observed by SCLIM. Representative 3D images of cells with (lower panels) or without (upper) NAA are shown. Dashed lines indicate the edge of cells. Right montages show 3D time-lapse images of the indicated areas. (D) Wild-type, Ret1-aid and Sec21-aid cells with NAA expressing Mnn9–mCherry (cis, magenta) and Gnt1–GFP (medial, green) were observed. Left panels show representative 3D images of cells. Dashed lines indicate the edge of cells. Right montages show 3D time-lapse images of the indicated areas. (E) Magnified images of selected time points in D are shown. Scale bars: 1 µm. (F) Bar graph shows the rate of maturation of cisternae from cis to medial. Error bars represent s.d. from at least five independent cells.

Open in a separate window

Fig. 6.

Depletion of COPI protein inhibits cis to trans cisternal maturation. Ret1-aid (A) and Sec21-aid (B) cells expressing Mnn9–mCherry (cis, magenta) and Gnt1–GFP (medial, green) were grown to a mid-logarithmic phase in synthetic medium with or without 1 mM NAA at 30°C and observed by SCLIM. Representative 3D images of cells with (lower panels) or without (upper) NAA are shown. Dashed lines indicate the edge of cells. Right panels show 3D time-lapse images of the indicated areas. Scale bars: 1 µm. Representative images from at least five independent cells are shown.

Domains are formed and maintained on Golgi cisternae during cisternal maturation

During cisternal maturation, segregation of early and late Golgi-resident proteins was observed in individual cisternae in our previous report (Matsuura-Tokita et al., 2006). Here, our 4D imaging with higher spatiotemporal resolutions reveals details of the compositional change in maturing individual cisternae (see Fig. 2). At the beginning of the transition of cis cisternae, marked by Mnn9–mCherry, Gnt1–GFP coming from medial cisternae began to localize to small regions on the cis cisternae. These small puncta gradually expanded their area without extensive mixing with the cis component and eventually covered the whole cisternae. Similar events were also observed for the cases of Sys1–GFP and Sec7–GFP coming into the cis-cisternae. Furthermore, two cis-Golgi marker proteins, Sed5 and Mnn9, showed segregated distribution on a single cisterna (Fig. 1C). Such segregation of membrane proteins might be an intrinsic property of the Golgi membrane and would be a key to understand their sorting and dynamic equilibrium. In COPI-depleted cells, domain segregation was still observed but in a fixed fashion (see Figs 5 and ​and6),6), suggesting the involvement of COPI in dynamic mixing of domains. The Rab family of GTPases regulate a variety of steps of membrane trafficking. They show phase-specific localization on the Golgi membrane, which is coordinately controlled by Rab GEF and Rab GAP cascades (Rivera-Molina and Novick, 2009; Suda et al., 2013). We previously showed that the cells disrupted for the Rab GAP cascade failed in proper Rab GTPase transition and thus in proper membrane segregation but somehow continued cisternal maturation (Suda et al., 2013). The significance of membrane domain segregation will need further investigation.

Microscopy

Fluorescence microscopy was performed by super-resolution confocal microscopy (SCLIM). The system setup had an Olympus model IX-71 inverted fluorescence microscope with a UPlanSApo 100×NA 1.4 oil objective lens (Olympus, Japan), a high-speed spinning-disk confocal scanner (Yokogawa Electric, Japan), a custom-made spectroscopic unit, image intensifiers (Hamamatsu Photonics, Japan) with a custom-made cooling system, and two EM-CCD cameras (Hamamatsu Photonics, Japan) for green and red channels (Kurokawa et al., 2013). For microscopic observation, all strains were grown in selective medium (0.67% yeast nitrogen base without amino acids and 2% glucose) with appropriate supplements. ret1-1 mutant cells were cultured at the permissive temperature (25°C) and then incubated at the restrictive temperature (38°C) for 10 min on a thermo-controlled stage (Tokai Hit, Japan) whose temperature was kept at 38°C. Ret1-aid cells and Sec21-aid cells were grown with 1 mM NAA for 2 h. For 4D live imaging of wild-type cells, Ret1-aid cells and Sec21-aid cells, 32 optical slices spaced 0.1 µm apart were sequentially collected at 8 frames per second (fps), which takes 5.7 s in total including the time for data transfer. For 4D observation of ret1-1 mutant cells, 32 optical slices 0.1 µm apart were collected at 15 fps every 5 s. z-stack images were converted to 3D voxel data and processed by iterative deconvolution with Volocity (Perkin Elmer) using a theoretical point-spread function for spinning-disc confocal microscopy. Maximum intensity projection images from these deconvolved 3D images were used for the calculation of the relative fluorescent values of the green and red signals of the Golgi areas with Fiji software (Schindelin et al., 2012). The center of mass of each cisterna was calculated in a mask of the cisterna from raw 4D images by Fiji software. The projection of the center of mass was conducted by Temporal-Color Code of Fiji plugin with a thermal lookup table.

We thank all members of the A.N. laboratory for helpful comments. We thank Dr Ryogo Hirata of Chemical Genetic Laboratory of RIKEN for valuable discussions and help.